Low-dose erythropoietin in anemic arthroplasty patients within contemporary patient blood management pathways

Abstract

Background

Preoperative anemia remains a major determinant of transfusion risk in total joint arthroplasty (TJA), even within contemporary patient blood management (PBM) pathways. Although erythropoietin (EPO) increases preoperative hemoglobin (Hb) levels, its role in modern low-transfusion arthroplasty practice remains debated. This study evaluated the real-world effectiveness of a selective, low-dose preoperative EPO protocol in anemic patients undergoing TJA within a comprehensive PBM program.

Methods

We retrospectively analyzed 2462 primary TJA procedures performed between 2013 and 2024 and identified 188 patients with baseline Hb < 12 g/dL. Of these, 80 received low-dose EPO and 108 served as controls. The EPO regimen consisted of two weekly subcutaneous injections of 20,000 IU administered preoperatively, combined with oral iron supplementation. All patients were managed under a standardized PBM protocol. The primary outcome was the bleeding index at postoperative day 7 (BI-7), defined as the sum of Hb decline between admission and postoperative day 7 and the number of transfused red-blood-cell units.

Results

Patients receiving EPO had lower baseline Hb levels than controls (10.9 vs. 11.5 g/dL). EPO treatment increased preoperative Hb by a mean of 1.3 g/dL and resulted in higher Hb levels at hospital admission, postoperative day 1, and postoperative day 7. No blood transfusions were administered during the observation period. BI-7 values were low and comparable between groups (2.1 vs. 2.3; p = 0.30). No thromboembolic events were observed.

Conclusions

Within a contemporary PBM framework, selective low-dose preoperative EPO safely optimized perioperative Hb levels and was associated with a transfusion-free perioperative course despite higher baseline hematologic risk. These findings suggest that modern low-transfusion PBM strategies can be effectively extended to selected anemic arthroplasty patients by improving preoperative Hb reserve.

Introduction

Primary total hip and knee arthroplasty procedures are frequently accompanied by perioperative blood loss and may necessitate allogeneic blood transfusions. Such transfusions carry well-recognized risks, including infection, transfusion-related acute lung injury, immunomodulation, and elevated postoperative morbidity [1–3]. Although patient blood management (PBM) programs, incorporating tranexamic acid, restrictive transfusion thresholds, and improved surgical methods, have substantially reduced transfusion rates [4, 5], preoperative hemoglobin (Hb) levels remain the strongest predictor of transfusion in total joint arthroplasty (TJA) [1, 6, 7].

Preoperative anemia affects 17–44% of patients undergoing arthroplasty [1, 8, 9]. Its management is challenging, largely due to variable responses to iron supplementation and persistent concerns regarding the safety, cost, and optimal dosing of erythropoietin (EPO) [10–12]. While EPO administration is known to increase preoperative Hb levels, it also widens the margin between postoperative Hb values and transfusion thresholds.

The present study investigates the real-world effectiveness and safety of a selective, low-dose preoperative EPO protocol in anemic patients undergoing primary TJA within an established PBM program [13, 14].

Methods

This retrospective registry-based study included consecutive adult patients who underwent primary unilateral total hip or knee arthroplasty between 2013 and 2024 and presented with baseline anemia. All procedures were performed by a single surgeon, and data were extracted from a prospectively maintained registry. Exclusion criteria were revision arthroplasty, bilateral procedures within three months, surgery for infection, tumor, or acute fracture, incomplete hematologic data, severe thrombocytopenia (platelet count < 50 × 10⁹/L), and concomitant non-arthroplasty procedures.

Collected variables included patient demographics, clinical parameters, and perioperative data. Preoperative variables were age, sex, body mass index (BMI), American Society of Anesthesiologists (ASA) classification [15], chronic use of antithrombotic agents, baseline Hb, and estimated glomerular filtration rate (eGFR). Intraoperative variables included tranexamic acid (TXA) administration and operative time. Postoperative variables included Hb concentrations and red blood cell (RBC) transfusion events.

All patients were managed under a standardized PBM program comprising TXA administration and restrictive transfusion thresholds. No postoperative wound drains were used for total hip arthroplasty; a superficial (extra-articular) drain was routinely placed following total knee arthroplasty. This PBM framework was established in 2005 and subsequently refined to enhance safety and cost-effectiveness [14, 16]. EPO was prescribed by the operating surgeon based on predefined clinical criteria within the PBM program, not a formal scoring system. Indications included Hb < 12 g/dL, advanced age, low body weight, female sex, cardiovascular comorbidities, and anticipated surgical complexity (e.g., complex hip reconstruction). Treatment decisions reflected individualized clinical judgment within this structured framework. In the absence of contraindications (EPO hypersensitivity, uncontrolled hypertension, or nutritional anemia), patients received subcutaneous epoetin alfa (Binocrit^®, Sandoz AG, Basel, Switzerland) at a dose of 20,000 IU once weekly for two weeks, combined with oral iron supplementation (Tardyferon^®, Pierre Fabre, France; 80 mg three times daily). The final EPO injection was administered 7 days prior to surgery. The control group did not receive systematic oral iron supplementation. This low-dose regimen was part of a surgeon-led selective PBM strategy.

All procedures used contemporary implants and standardized surgical techniques. Anticoagulant and antiplatelet therapies were managed according to institutional protocols, typically involving temporary perioperative discontinuation to minimize bleeding risk [17]. Postoperative thromboprophylaxis consisted of rivaroxaban 10 mg daily.

Hb was measured at four predefined time points: baseline (within three months preoperatively), the day before surgery, postoperative day (POD) 1, and POD-7 (± 2 days). Patients discharged before POD-5 underwent outpatient laboratory testing for POD-7 Hb assessment. All allogeneic RBC transfusions administered from the preoperative day through POD-7 were recorded.

The primary outcome was the bleeding index at POD-7 (BI-7), calculated as the decline in Hb (g/dL) from hospital admission to POD-7 (± 2 days), plus the number of RBC units transfused during this interval. Each transfused RBC unit was assumed to increase Hb by approximately 1 g/dL [18, 19].

Secondary outcomes included allogeneic RBC transfusion within the first postoperative week and postoperative complications occurring within 60 days. Severe postoperative anemia was defined as Hb < 8.0 g/dL at POD-7.

Baseline characteristics were summarized using means ± standard deviation (SD) for continuous variables and counts with percentages for categorical variables. Comparisons between EPO and non-EPO groups in the unmatched cohort were conducted using independent t-tests or Pearson’s chi-square tests, as appropriate. To account for nonrandom treatment allocation, multiple ordinary least squares (OLS) regression was employed to assess the association between EPO treatment and BI-7. The final prespecified OLS model included EPO treatment status as the exposure of interest and was adjusted for a comprehensive set of potential confounders: age, BMI, baseline Hb, renal function (Modification of Diet in Renal Disease eGFR), sex, ASA class, joint type (hip vs. knee arthroplasty), and prior antithrombotic therapy. Model assumptions were verified, and robust (Huber–White) standard errors were applied to address potential heteroskedasticity. Statistical analyses were performed using Stata version 15.1 (StataCorp, College Station, TX, USA). A two-sided p-value of < 0.05 was considered statistically significant.

Results

Of the 2462 primary TJA procedures identified during the study period, 32 were excluded because of incomplete data (n = 26), nonelective indications (n = 4), or combined procedures (n = 2). A total of 2430 procedures were therefore eligible for analysis (Fig. 1). Of these, 1556 (64%) were total hip arthroplasties and 874 (36%) were total knee arthroplasties.

Fig. 1.

Study flow chart

Among the 2,430 eligible procedures, 188 patients (6.2%) had baseline Hb levels < 12 g/dL. Within this subgroup, 80 patients (43%) received EPO therapy. Targeted administration of EPO was evident, as 83% of patients with baseline Hb < 11 g/dL received EPO treatment. All patients completed the prescribed EPO regimen except one, whose treatment was interrupted due to logistical constraints during the COVID-19 pandemic.

Baseline characteristics of patients with baseline anemia are presented in Table 1. EPO recipients had significantly lower baseline Hb (10.9 vs. 11.5 g/dL; p < 0.001), lower eGFR (61 vs. 74 mL/min; p = 0.003), and were slightly older (77.9 vs. 75.3 years; p = 0.047) than the non-EPO group. Other baseline characteristics—sex distribution, BMI, ASA class, joint replaced, and preoperative antithrombotic therapy—were similar between groups.

Table 1.

Baseline characteristics of patients with baseline Hb < 12 g/dL

Characteristics	Non-EPO (n = 108)	EPO (n = 80)	P-value
Age (years), mean ± SD	75.3 ± 8.9	77.9 ± 8.7	0.047
BMI (kg/m²), mean ± SD	30.8 ± 6.8	29.1 ± 6.6	0.093
Baseline Hb (g/dL), mean ± SD	11.5 ± 0.4	10.9 ± 0.8	< 0.001
Preoperative Hb (g/dL), mean ± SD	11.8 ± 0.8	12.2 ± 1.0	< 0.001
eGFR (MDRD, mL/min/1.73 m²), mean ± SD	73.7 ± 28.6	61.4 ± 23.7	0.003
MCV femtoliters (fL), mean ± SD	93.6 ± 6.9	91.6 ± 6.3	0.052
Surgical time (min), mean ± SD	66.6 ± 13.2	64.4 ± 14.3	0.262
Female sex, N (%)	87 (80.6%)	61 (76.3%)	0.478
Procedure type (TKA), n (%)	45 (41.7%)	28 (35.0%)	0.354
ASA classification, n (%)			0.586
ASA 1	4 (3.7%)	3 (3.8%)
ASA 2	75 (69.4%)	50 (62.5%)
ASA 3	29 (26.9%)	27 (33.8%)
Prior antithrombotic use, n (%)	49 (45.8%)	41 (51.9%)	0.459
TXA use, N (%)	101 (95.3%)	71 (93.4%)	0.744

Abbreviations—SD: Standard Deviation; BMI: Body Mass Index; Hb: Hemoglobin; eGFR: Estimated Glomerular Filtration Rate; MDRD: Modification of Diet in Renal Disease; MCV: Mean Corpuscular Volume; TKA: Total Knee Arthroplasty; ASA: American Society of Anesthesiologists; TXA: Tranexamic Acid.

The unadjusted perioperative Hb trajectory across four predefined time points is presented in Table 2; Fig. 2. EPO administration increased preoperative mean Hb by 1.3 g/dL, resulting in higher mean Hb levels at hospital admission, POD-1, and POD-7 compared with controls.

Table 2.

Unadjusted clinical outcomes

Outcome	Non-EPO (n = 108)	EPO (n = 80)	P-value
Hb Day 1 (g/dL), mean ± SD	10.8 ± 0.8	11.2 ± 1.1	0.009
Hb Day 7 (g/dL), mean ± SD	9.5 ± 1.0	10.2 ± 1.1	< 0.001
BI, mean ± SD	2.3 ± 1.0	2.1 ± 1.1	0.202
RBC transfusion between the day before surgery and POD-7	1 (0.9%)	0 (0.0%)	1.00

Fig. 2.

Unadjusted hemoglobin (Hb) trajectory. Unadjusted perioperative Hb (g/dL) and 95% confidence intervals for EPO and non-EPO groups. Preoperative Hb refers to the value at admission (one day before surgery). Hb Day 1 is the Hb level measured on the first postoperative day

Abbreviations—SD: Standard Deviation; Hb: Hemoglobin; BI: Bleeding Index; RBC: Red Blood Cell; POD: Postoperative Day.

The unadjusted bleeding index at POD-7 was slightly lower in the EPO group (2.1 ± 1.1) than in controls (2.3 ± 1.0), corresponding to a crude mean difference of − 0.2 (95% CI − 0.5 to + 0.1; p = 0.20). After multivariable adjustment, EPO treatment was not significantly associated with BI-7 (β = − 0.20; 95% CI − 0.58 to + 0.18; p = 0.30). Other covariates—including age, sex, ASA class, joint type, and renal function—were likewise not significantly associated with the bleeding index (Table 3).

Table 3.

Multivariable regression of bleeding index (Day 7)

Variable	Coefficient b	Robust SE	95% CI	P-value
EPO (yes or no)	− 0.20	0.19	− 0.58 to + 0.18	0.299
Age (centered, per year)	− 0.013	0.013	− 0.037 to + 0.012	0.322
BMI (centered, per kg/m²)	− 0.024	0.013	− 0.050 to + 0.003	0.082
Baseline Hb (centered, per g/dL)	− 0.004	0.019	− 0.041 to + 0.033	0.811
eGFR (MDRD, centered)	0.004	0.003	− 0.002 to + 0.011	0.200
Male sex	− 0.14	0.21	− 0.56 to + 0.27	0.495
ASA II vs. I	0.22	0.49	− 0.75 to + 1.19	0.654
ASA III vs. I	0.42	0.53	− 0.62 to + 1.46	0.428
TKA vs. THA	0.09	0.18	− 0.27 to + 0.44	0.642
Prior antithrombotic therapy	− 0.30	0.18	− 0.64 to + 0.05	0.094
Intercept	2.15	0.48	+ 1.20 to + 3.10	< 0.001

Abbreviations—BI-7: Bleeding Index at Postoperative Day 7; EPO: Erythropoietin; BMI: Body Mass Index; eGFR (MDRD): Estimated Glomerular Filtration Rate, calculated using the Modification of Diet in Renal Disease equation; ASA: American Society of Anesthesiologists Physical Status Classification; TKA/THA: Total Knee Arthroplasty/Total Hip Arthroplasty.

Centered variables: continuous variables were mean-centered prior to analysis. Individual preoperative hemoglobin responses to EPO are shown in Fig. 3. The majority of patients demonstrated a clinically meaningful hemoglobin increase, with a trend toward larger responses at lower baseline hemoglobin levels. A minority of patients, predominantly those with milder baseline anemia, showed minimal or no preoperative hemoglobin change; however, BI-7 values in these patients remained low.

Fig. 3.

Individual preoperative hemoglobin response in EPO-treated patients. Each point represents one patient. The x-axis shows baseline hemoglobin (g/dL); the y-axis shows the change in hemoglobin from baseline to the preoperative value. The dashed line represents a linear fit. The horizontal reference line indicates no change

No blood transfusions were administered during the observation period (admission to POD 7). Severe postoperative anemia (Hb < 8.0 g/dL) occurred in 2.5% of EPO-treated patients and 7.4% of controls (p = 0.19). No thromboembolic events occurred. One postoperative death occurred in the EPO group at a rehabilitation facility and was adjudicated as unrelated to treatment.

Discussion

This study evaluated the effectiveness and safety of a selective, low-dose EPO protocol in anemic patients undergoing primary TJA within a contemporary PBM framework. To the best of our knowledge, this is the first investigation to assess a regimen consisting of two preoperative injections of 20,000 IU EPO integrated into a modern PBM pathway, with the final EPO injection administered 7 days before surgery.

Historically, recombinant human EPO has been administered using manufacturer-recommended high-dose protocols, typically involving weekly injections of 600 IU/kg for three weeks or daily injections of 300 IU/kg over a 14-day perioperative period. While these regimens reduce allogeneic transfusion requirements, they are associated with increased cost, logistical complexity, and concerns related to supraphysiologic Hb targets and thromboembolic risk [10–12, 20]. Consequently, the minimum effective EPO dose required for clinically meaningful Hb optimization in contemporary arthroplasty remains uncertain.

Previous studies have suggested that lower EPO doses may be sufficient in selected patients. Rosencher et al. demonstrated that one or two injections of 40,000 IU were often adequate to achieve target Hb levels [21], while Feagan et al. reported comparable Hb increases with 20,000 IU and 40,000 IU regimens [22]. These findings support the rationale for proportionate, lower-dose EPO strategies [20].

The present study extends this evidence by demonstrating that two injections of 20,000 IU EPO can be effectively implemented in routine clinical practice. Notably, no patient receiving EPO required allogeneic blood transfusion, despite significantly lower baseline Hb levels than controls. These findings indicate that selective low-dose EPO can mitigate transfusion risk in higher-risk patients and support its role as a complementary adjunct within optimized PBM programs.

The bleeding index at POD-7 (BI-7) was low in both EPO-treated and control groups, reflecting the effectiveness of contemporary PBM practices aimed at minimizing perioperative blood loss [23, 24]. Additionally, reduced whole-blood viscosity and lower red-cell mass in anemic patients may contribute to lower absolute red-cell loss. Together, these factors likely explain the limited differences in red-cell loss between the two groups.

Although no significant difference in BI-7 was observed, EPO-treated patients consistently achieved higher perioperative Hb levels, resulting in a wider margin between postoperative Hb values and transfusion thresholds. In TXA-based PBM settings, where baseline transfusion rates are already low, incremental reductions in transfusion risk are inherently difficult to detect. Nevertheless, achieving zero transfusions in a higher-risk anemic population remains clinically meaningful.

The absence of an association between EPO use and BI-7 is consistent with the pharmacologic profile of epoetin alfa. In our protocol, the final EPO injection was administered 7 days before surgery, such that circulating epoetin is expected to be largely cleared at the time of surgery, while erythropoietic effects induced preoperatively may persist into the early postoperative period. EPO enhances erythropoiesis and increases red-cell mass but does not influence surgical blood loss. Accordingly, postoperative Hb decline was similar between the groups, while absolute Hb levels were consistently higher in EPO-treated patients. Importantly, Hb values in this cohort remained above 10 g/dL at POD-7, a threshold associated with reduced cardiorespiratory stress and safer recovery [25].

Renal impairment was more common among EPO recipients. Given that early renal dysfunction is associated with relative endogenous EPO deficiency, the consistent Hb response observed suggests that low-dose EPO effectively addresses this mechanism. The combination of improved preoperative Hb and elimination of transfusion further supports the role of selective low-dose EPO as a complementary PBM intervention.

From a practical perspective, the standardized 20,000 IU dosing schedule enabled timely treatment and excellent adherence within a surgeon-led workflow. The reduced-dose approach limited costs and avoided the complexity of traditional high-dose regimens, while the use of oral rather than intravenous iron enhanced feasibility. Incorporation of the BI-7 metric, a physiologically grounded measure of postoperative red-cell loss, strengthened the interpretability of these findings.

No thromboembolic events were observed among EPO-treated patients. This favorable safety profile aligns with prior literature and likely reflects low-dose administration, avoidance of supraphysiologic Hb targets, and standardized thromboprophylaxis [22, 26–28].

This study has several limitations. Its retrospective, single-center design and nonrandomized treatment allocation limit causal inference and introduce potential residual confounding. Iron parameters (ferritin, transferrin saturation) were not systematically collected, precluding differentiation between iron-deficiency anemia and anemia of chronic disease, and limiting the ability to predict individual EPO responsiveness. Although hemoglobin and platelet counts were assessed at admission and severe thrombocytopenia (platelet count < 50 × 10⁹/L) was a predefined exclusion criterion, platelet values were not systematically recorded in the registry during the study period and were therefore not included in the analysis. This is unlikely to have materially affected interpretation, as clinically significant thrombocytopenia is uncommon in elective arthroplasty, patients with coagulation disorders were not candidates for surgery, and the study focused on red-cell loss rather than coagulation parameters. Furthermore, the observed preoperative hemoglobin increase reflects the combined effect of EPO and oral iron supplementation, as oral iron was systematically co-administered only to EPO-treated patients; the independent contribution of EPO cannot therefore be isolated. In addition, the very low transfusion event rate inherent to contemporary PBM pathways restricts statistical power to detect differences in transfusion-related outcomes.

Conclusion

Selective low-dose preoperative EPO safely increased perioperative Hb levels in anemic patients undergoing primary TJA, supporting the extension of contemporary low-transfusion PBM strategies to selected high-risk patients.

Author contributions

HH designed the study, implemented the clinical protocol, collected the data, and prepared the initial manuscript draft. PF performed the statistical analysis. HH, PF, and GF contributed to manuscript editing and revision. All authors contributed to data interpretation and approved the final version.

Funding

This research received no external funding.

Data availability

The datasets supporting the conclusions of this article are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The study did not alter standard clinical practice and was conducted in accordance with French law. In accordance with French regulations governing retrospective analyses of anonymized routinely collected clinical data (Code de la santé publique), formal approval by an institutional review board or ethics committee was not required. The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients prior to surgery.

Consent for publication

Not applicable.

Conflict of interests

The authors declare no competing interests.